CN106897545B - A kind of tumor prognosis forecasting system based on depth confidence network - Google Patents
A kind of tumor prognosis forecasting system based on depth confidence network Download PDFInfo
- Publication number
- CN106897545B CN106897545B CN201710007736.5A CN201710007736A CN106897545B CN 106897545 B CN106897545 B CN 106897545B CN 201710007736 A CN201710007736 A CN 201710007736A CN 106897545 B CN106897545 B CN 106897545B
- Authority
- CN
- China
- Prior art keywords
- patient
- data
- variable
- classification
- hidden layer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 206010028980 Neoplasm Diseases 0.000 title claims abstract description 41
- 238000004393 prognosis Methods 0.000 title claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000013135 deep learning Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000003993 interaction Effects 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 17
- 238000000034 method Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 7
- 230000004083 survival effect Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 201000011510 cancer Diseases 0.000 description 5
- 201000010099 disease Diseases 0.000 description 5
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 5
- 230000036962 time dependent Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 206010027476 Metastases Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 210000001165 lymph node Anatomy 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
- 230000001173 tumoral effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of tumor prognosis forecasting systems based on depth confidence network, comprising: for acquiring the data acquisition module of tumor information；For carrying out the data preprocessing module of missing values processing and normalized to tumour initial data；Data for carrying out deep learning and prediction modeling to tumour data learn prediction module；Prediction result display module for showing the relative risk of data study prediction module output；The present invention is limited Boltzmann machine using Gauss, retains the nonlinear characteristic of data；It, can be with flexible expansion depth confidence network according to the accuracy of the dimension of input data, the quantity of output category, model；During model training, any restrictions are not used and it is assumed that variable can sufficiently be excavated to the interaction between the influence mode and variable of result, show the mode that different factors influence tumor prognosis comprehensively, and improve the accuracy of tumor prognosis prediction.
Description
Technical field
The present invention relates to tumour forecasting system more particularly to a kind of tumor prognosis forecasting systems based on depth confidence network
Background technique
The morbidity and mortality of cancer are high, have become the main reason for mankind are because of disease death.With the size of population
Growth and aging of population development, cancer bring Disease Spectrum further increases, and becomes current medical expense branch
Most paid.Tumor prognosis forecast analysis can provide the prognosis information for being used for disease treatment to clinician, help to control
The formulation for the treatment of scheme improves disease cured rate, improves patient's prognosis quality of life, Disease Spectrum is effectively reduced, for cancer
Control and therapeutic potential are great.American cancer federation release based on tumor invasive depth, lymph node, DISTANT METASTASES IN TNM
In the cancer clinical practice of Staging System worldwide, it is widely applied, is the weight of guiding treatment and clinical research
Want tool.But it is many newest the study found that TNM stage cannot existence final result difference to the patient of different tumoral characters
It distinguishes.Neural network can sufficiently approach arbitrarily complicated non-linear relation, study and adaptive unknown and not really with it
Fixed system, with robustness and fault-tolerance, qualitatively and quantitatively data can be handled simultaneously and can large-scale parallel distribution process
Advantage is used widely in tumor prognosis forecast analysis.
Generally existing censored data in tumor prognosis data, censored data are not missing data, but are only capable of providing starting point
To the prognosis information of truncated time, the deficiency of data for the complete information that starting point occurs to event cannot be provided.It is existing to be based on
The tumor prognosis prediction analysis method of neural network, or censored data cannot be made full use of；Or making full use of censored data
In the case of, it not can effectively solve the Time Dependent and nonlinear problem of Prognostic Factors；Or obtained survivorship curve is not in monotonicity；
Or constructed neural network does not have scalability, is unfavorable for the large scale processing of mass data.
Deep learning is the popular domain of current machine Learning Studies, because it is with autonomous feature learning ability and high precision
Property be applied to many fields, including speech recognition, image procossing, natural language processing and personage's portrait etc., but current depth
Study is also seldom applied to tumor prognosis forecast analysis field.
Summary of the invention
In view of the above-mentioned deficiencies in the prior art, it is an object of the present invention to provide a kind of tumor prognosis based on depth confidence network
Forecasting system improves prediction analysis method neural network based using the depth confidence network algorithm in deep learning field,
In the case where making full use of censored data, the Time Dependent and nonlinear problem of Prognostic Factors are effectively solved, improves tumor prognosis
The accuracy of prediction, auxiliary doctor formulate the treatment plan of patient；The depth confidence network of building is with good expansibility,
Conducive to the large scale processing of mass data.
The purpose of the present invention is achieved through the following technical solutions: a kind of tumor prognosis based on depth confidence network
Forecasting system, the system include: the data acquisition module for acquiring tumor information；For being lacked to tumour initial data
The data preprocessing module of value processing and normalized；For tumour data to be carried out with the data of deep learning and prediction modeling
Learn prediction module；Prediction result display module for showing the relative risk of data study prediction module output；
The treatment process of the data study prediction module includes two parts: being primarily based on the unsupervised training method of deep learning, benefit
Similar patients are clustered with patient characteristic, secondly utilize similar patients group, calculate accumulative risk function, specific steps are such as
Under:
(1) similar patients are clustered using depth confidence network model
(1.1) assume that patient populations are n, patient characteristic quantity is m, it is seen that layer variable is vi, i=1 ..., m, hidden layer
Variable is hj, j=1 ..., g, wherein m is the quantity of visible layer variable, and g is the quantity of hidden layer variable；wijIt is visible layer variable
viWith hidden layer variable hjBetween connection weight, then, it is seen that connection weight matrix W=(w between layer and hidden layerij)m×g；It can
See the biasing a=(a of layer variable1,…,am), the biasing b=of hidden layer variable (1,…,bg)。
(1.2) Gauss RBM model is constructed: because the characteristic of patient is often some continuous variables or orderly becomes
Therefore amount, rather than simple two-category data replace simple RBM model using Gauss RBM model, to retain data
Nonlinear characteristic.Energy function E (v, the h of Gauss RBM model；θ) are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model, σ=(σ1,…,σm) indicate visible layer variable Gauss
Noise.The condition of Gauss RBM is distributed are as follows:
Wherein, N (μ, σ2) expression mean value be μ, standard deviation be σ Gaussian Profile.The edge distribution of visible layer v are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model.Using gradient descent method adjusting parameter, make input with it is defeated
Error out is minimum, by meeting following formula, obtains the optimal parameter of model:
Wherein, Z (θ) is normaliztion constant.
(1.3) in training, depth confidence network is using successively unsupervised method come learning parameter.First visible layer
V and hidden layer h1 trains the parameter W of this RBM as a limited Boltzmann machine RBM1；Then, W is kept1It is constant,
H1 trains the parameter W of second RBM using h2 as hidden layer as a visible layer2；Then, W is kept2It is constant, h2 is made
The parameter W of third RBM is trained using h3 as hidden layer for a visible layer3；And so on, it trains complicated by more
The depth confidence network that layer RBM is stacked.In training process, the interaction between dominated variable and variable be not to classification results
Influence form.
(1.4) because the value of hidden layer variable is two-value type data, all hidden layers that we can use top become
The value of amount is classified for one belonging to patient to determine；If top has n hidden layer variable, patient is divided into 2n
Class.
(1.5) increase patient populations, have no need to change network settings；Increase patient characteristic, increases visible layer in a network
The quantity of variable；Patient classification's quantity is adjusted, modifies the variable quantity of top hidden layer in a network；Adjust the accurate of model
Degree, thus it is possible to vary the number of plies of hidden layer.
(2) similar patients group is utilized, calculate accumulative risk function: patient i has m input feature vector, is denoted as Xi, in step
Patient i obtains unique classification c, c ∈ P in 1；P is to own using depth confidence network model to what similar patients clustered
Category set；In time t, and the accumulative risk function H of patient i (t | Xi) be exactly c classification Nelson-Aalen estimated value:
Wherein, dl,cIt indicates in time tl,c, the death toll of patient in c classification；rl,cIt indicates in time tl,c, in c classification
Patient there are the numbers of risk；t1,c< t2,c< ... < tN(c),cIndicate a different event time of N (c) in c classification；
(T1,c,s1,c),…,(Tn(c),c,sn(c),c) indicate c classification in all patients life span and survival condition, n
(c) total quantity of c class patient is indicated；To a patient i, if si,c=0, then the patient is in time Ti,cBelong to censored data
(survival or lost to follow-up)；If si,c=1, then the patient is in time Ti,cOccur result event (death)；Patient i is in time tl,c's
Survival conditionWherein I () is indicator function, works as Ti,c< tl,cWhen,Work as Ti,c
≥tl,cWhen,Then in time tl,c, there are the number r of risk in c classificationl,c=rl-1,c-dl-1,c, death tollWherein r0,c=n (c), d0,c=0.
The beneficial effects of the present invention are:
1) it is limited Boltzmann machine using Gauss, retains the nonlinear characteristic of data；
It 2), can be with flexible expansion depth confidence according to the accuracy of the dimension of input data, the quantity of output category, model
Network；
3) during model training, any restrictions are not used and it is assumed that variable can sufficiently be excavated to the influence side of result
Interaction between formula and variable shows the mode that different factors influence tumor prognosis comprehensively, and improves tumor prognosis
The accuracy of prediction；
4) on the basis of clustering using depth confidence network to patient, principle is retained based on event, using Nelson-
Aalen estimation function calculates the accumulative risk function of patient, guarantees that monotonicity is presented in output survivorship curve.
Detailed description of the invention
Fig. 1 is that the present invention is based on the tumor prognosis forecasting system frame diagrams of depth confidence network；
Fig. 2 is the tumor prognostic analysis algorithm flow chart based on depth confidence network；
Fig. 3 is depth confidence network model.
Specific embodiment
Invention is further described in detail in the following with reference to the drawings and specific embodiments.
Censored data in the present invention are as follows: if the data for result event do not occur are referred to as in the defined end time
For censored data, the time from starting point to truncation is known as truncated time.Time Dependent phenomenon are as follows: no matter baseline risk,
It is constant there are the opposite risk there is no the individual generation event of the exposure of the individual of a certain exposure in any time point；
Prognostic Factors do not meet the phenomenon that above-mentioned hypothesis, and being regarded as influence of the Prognostic Factors to tumor prognosis, there are Time Dependents.
As shown in Figure 1, a kind of tumor prognosis forecasting system based on depth confidence network provided by the invention, comprising: use
In the data acquisition module of acquisition tumor information；For carrying out the number of missing values processing and normalized to tumour initial data
Data preprocess module；Data for carrying out deep learning and prediction modeling to tumour data learn prediction module；For that will count
The prediction result display module shown according to the relative risk of study prediction module output；The data study prediction module
Treatment process includes two parts: be primarily based on the unsupervised training method of deep learning, using patient characteristic to similar patients into
Row cluster, secondly utilizes similar patients group, calculates accumulative risk function, as shown in Figure 2, the specific steps are as follows:
(1) similar patients are clustered using depth confidence network model, depth confidence network model is as shown in Figure 3；
(1.1) assume that patient populations are n, patient characteristic quantity is m, it is seen that layer variable is vi, i=1 ..., m, hidden layer
Variable is hj, j=1 ..., g, wherein m is the quantity of visible layer variable, and g is the quantity of hidden layer variable；wijIt is visible layer variable
viWith hidden layer variable hjBetween connection weight, then, it is seen that connection weight matrix W=(w between layer and hidden layerij)m×g；It can
See the biasing a=(a of layer variable1,…,am), the biasing b=(b of hidden layer variable1,…,bg)。
(1.2) Gauss RBM model is constructed: because the characteristic of patient is often some continuous variables or orderly becomes
Therefore amount, rather than simple two-category data replace simple RBM model using Gauss RBM model, to retain data
Nonlinear characteristic.Energy function E (v, the h of Gauss RBM model；θ) are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model, σ=(σ1,…,σm) indicate visible layer variable Gauss
Noise.The condition of Gauss RBM is distributed are as follows:
Wherein, N (μ, σ2) expression mean value be μ, standard deviation be σ Gaussian Profile.The edge distribution of visible layer v are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model.Using gradient descent method adjusting parameter, make input with it is defeated
Error out is minimum, by meeting following formula, obtains the optimal parameter of model:
Wherein, Z (θ) is normaliztion constant.
(1.3) in training, depth confidence network is using successively unsupervised method come learning parameter.First visible layer
V and hidden layer h1 trains the parameter W of this RBM as a limited Boltzmann machine RBM1；Then, W is kept1It is constant,
H1 trains the parameter W of second RBM using h2 as hidden layer as a visible layer2；Then, W is kept2It is constant, h2 is made
The parameter W of third RBM is trained using h3 as hidden layer for a visible layer3；And so on, it trains complicated by more
The depth confidence network that layer RBM is stacked.In training process, the interaction between dominated variable and variable be not to classification results
Influence form.
(1.4) because the value of hidden layer variable is two-value type data, all hidden layers that we can use top become
The value of amount is classified for one belonging to patient to determine；If top has n hidden layer variable, patient is divided into 2n
Class.
(1.5) increase patient populations, have no need to change network settings；Increase patient characteristic, increases visible layer in a network
The quantity of variable；Patient classification's quantity is adjusted, modifies the variable quantity of top hidden layer in a network；Adjust the accurate of model
Degree, thus it is possible to vary the number of plies of hidden layer.
(2) similar patients group is utilized, calculate accumulative risk function: patient i has m input feature vector, is denoted as Xi, in step
Patient i obtains unique classification c, c ∈ P in 1；P is to own using depth confidence network model to what similar patients clustered
Category set；In time t, and the accumulative risk function H of patient i (t | Xi) be exactly c classification Nelson-Aalen estimated value:
Wherein, dl,cIt indicates in time tl,c, the death toll of patient in c classification；rl,cIt indicates in time tl,c, in c classification
Patient there are the numbers of risk；t1,c< t2,c< ... < tN(c),cIndicate a different event time of N (c) in c classification；
(T1,c,s1,c),…,(Tn(c),c,sn(c),c) indicate c classification in all patients life span and survival condition, n
(c) total quantity of c class patient is indicated；To a patient i, if si,c=0, then the patient is in time Ti,cBelong to censored data
(survival or lost to follow-up)；If si,c=1, then the patient is in time Ti,cOccur result event (death)；Patient i is in time tl,c's
Survival conditionWherein I () is indicator function, works as Ti,c< tl,cWhen,Work as Ti,c
≥tl,cWhen,Then in time tl,c, there are the number r of risk in c classificationl,c=rl-1,c-dl-1,c, death tollWherein r0,c=n (c), d0,c=0.The death rate of patient iWith dead
Rate is died, survivorship curve can be drawn.
The present invention utilizes the depth confidence network algorithm in deep learning field, improves forecast analysis side neural network based
Method effectively solves the Time Dependent and nonlinear problem of Prognostic Factors in the case where making full use of censored data, improves tumour
The accuracy of prognosis prediction, auxiliary doctor formulate the treatment plan of patient；Guarantee that monotonicity is presented in obtained survivorship curve simultaneously,
The depth confidence network of building is with good expansibility, conducive to the large scale processing of mass data.
Claims (1)
1. a kind of tumor prognosis forecasting system based on depth confidence network, which is characterized in that the system includes: swollen for acquiring
The data acquisition module of tumor information；For carrying out the data prediction of missing values processing and normalized to tumour initial data
Module；Data for carrying out deep learning and prediction modeling to tumour data learn prediction module；It is pre- for learning data
Survey the prediction result display module that the relative risk of module output is shown；The treatment process of the data study prediction module
Including two parts: it is primarily based on the unsupervised training method of deep learning, similar patients are clustered using patient characteristic,
It is secondary to utilize similar patients group, calculate accumulative risk function, the specific steps are as follows:
(1) similar patients are clustered using depth confidence network model
(1.1) assume that patient populations are N, patient characteristic quantity is M, it is seen that layer variable is vi, i=1 ..., m, hidden layer variable
For hj, j=1 ..., g, wherein m is the quantity of visible layer variable, and g is the quantity of hidden layer variable, it is seen that the quantity m of layer variable
Equal to patient characteristic quantity M；wijIt is visible layer variable viWith hidden layer variable hjBetween connection weight, then, it is seen that layer and hide
Connection weight matrix W=(w between layerij)m×g；Visible layer variable is biased to ai, i=1 ..., m, it is seen that layer variable it is inclined
It sets vector and is denoted as a=(a1..., am), hidden layer variable is biased to bj, j=1 ..., g, the bias vector of hidden layer variable
It is denoted as b=(b1..., bg)；
(1.2) Gauss RBM model is constructed: because the characteristic of patient is often some continuous variables or ordered set,
Rather than therefore simple two-category data replaces simple RBM model using Gauss RBM model, to retain the non-of data
Linear character；Energy function E (v, the h of Gauss RBM model；θ) are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model, σ=(σ1..., σm) indicate that the Gauss of visible layer variable makes an uproar
Sound；The condition of Gauss RBM is distributed are as follows:
Wherein, N (μ, p2) expression mean value be μ, standard deviation be p Gaussian Profile；The edge distribution of visible layer v are as follows:
Wherein, θ=(a, b, W, σ) indicates the setting parameter of model；Using gradient descent method adjusting parameter, make input and output
Error is minimum, by meeting following formula, obtains the optimal parameter of model:
Wherein, Z (θ) is normaliztion constant；
(1.3) in training, depth confidence network is using successively unsupervised method come learning parameter；First visible layer v and
Hidden layer h1 trains the parameter W of this RBM as a limited Boltzmann machine RBM1；Then, W is kept1It is constant, h1 is made
The parameter W of second RBM is trained using h2 as hidden layer for a visible layer2；Then, W is kept2It is constant, using h2 as one
A visible layer trains the parameter W of third RBM using h3 as hidden layer3；And so on, it trains complicated by multilayer
The depth confidence network that RBM is stacked；In training process, the not shadow of the interaction between dominated variable and variable to classification results
The form of sound；
(1.4) because the value of hidden layer variable is two-value type data, using the value of all hidden layer variables of top come really
Determine a classification belonging to patient；If top has n hidden layer variable, patient is divided into 2nClass；
(1.5) increase patient populations, have no need to change network settings；Increase patient characteristic, increases visible layer variable in a network
Quantity；Patient classification's quantity is adjusted, modifies the variable quantity of top hidden layer in a network；The accuracy of model is adjusted,
It can change the number of plies of hidden layer；
(2) similar patients group is utilized, calculate accumulative risk function: patient i has m input feature vector, is denoted as Xi, suffer from step 1
Person i obtains unique classification c, c ∈ P；P is all categories collection clustered using depth confidence network model to similar patients
It closes；In time t, and the accumulative risk function H of patient i (t | Xi) be exactly c classification Nelson-Aalen estimated value:
Wherein, dL, cIt indicates in time tL, c, the death toll of patient in c classification；rL, cIt indicates in time tL, c, patient in c classification
There are the numbers of risk；t1, c< t2, c< ... < tN (c), cIndicate a different event time of N (c) in c classification；
(T1, c, s1, c) ..., (TN (c), c, sN (c), c) indicate c classification in all patients life span and survival condition, n (c) table
Show the total quantity of c class patient；To a patient i, if sI, c=0, then the patient is in time TI, cBelong to censored data；If
sI, c=1, then the patient is in time TI, cThere is result event；Remember that patient i is in time t in c classificationL, cSurvival condition beThenWherein I () is indicator function, works as TI, c< tL, cWhen,Work as TI, c
≥tL, cWhen,Then in time tL, c, there are the number r of risk in c classificationL, c=rL-1, c-dL-1, c, death tollWherein r0, c=n (c), d0, c=0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710007736.5A CN106897545B (en) | 2017-01-05 | 2017-01-05 | A kind of tumor prognosis forecasting system based on depth confidence network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710007736.5A CN106897545B (en) | 2017-01-05 | 2017-01-05 | A kind of tumor prognosis forecasting system based on depth confidence network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106897545A CN106897545A (en) | 2017-06-27 |
| CN106897545B true CN106897545B (en) | 2019-04-30 |
Family
ID=59198346
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710007736.5A Active CN106897545B (en) | 2017-01-05 | 2017-01-05 | A kind of tumor prognosis forecasting system based on depth confidence network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106897545B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111184948A (en) * | 2020-01-09 | 2020-05-22 | 北京理工大学 | Vascular targeted photodynamic therapy-based nevus flammeus treatment method and system |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107491638A (en) * | 2017-07-28 | 2017-12-19 | 深圳和而泰智能控制股份有限公司 | A kind of ICU user's prognosis method and terminal device based on deep learning model |
| CN108428478B (en) * | 2018-02-27 | 2022-03-29 | 东北师范大学 | Thyroid cancer risk prediction method based on heterogeneous medical data mining |
| CN108959841A (en) * | 2018-04-16 | 2018-12-07 | 华南农业大学 | A kind of drug targeting albumen effect prediction technique based on DBN algorithm |
| JP7059162B2 (en) | 2018-10-29 | 2022-04-25 | 株式会社日立製作所 | Analytical instruments, analytical methods, and analytical programs |
| CN112289442B (en) * | 2018-10-29 | 2024-05-03 | 南京医基云医疗数据研究院有限公司 | Method and device for predicting disease end point event and electronic equipment |
| CN110348241B (en) * | 2019-07-12 | 2021-08-03 | 之江实验室 | Multi-center collaborative prognosis prediction system under data sharing strategy |
| CN111261299B (en) * | 2020-01-14 | 2022-02-22 | 之江实验室 | Multi-center collaborative cancer prognosis prediction system based on multi-source transfer learning |
| CN111640510A (en) * | 2020-04-09 | 2020-09-08 | 之江实验室 | Disease prognosis prediction system based on deep semi-supervised multitask learning survival analysis |
| CN111612278A (en) * | 2020-06-01 | 2020-09-01 | 戴松世 | Life state prediction method and device, electronic equipment and storage medium |
| CN114417956A (en) * | 2021-12-02 | 2022-04-29 | 中国人民解放军陆军军医大学第一附属医院 | Prognostic prediction model modeling method based on deep learning and interpretable mechanism |
| CN114171199A (en) * | 2021-12-14 | 2022-03-11 | 中国科学院深圳先进技术研究院 | Survival prediction method, system, terminal and storage medium for brain malignant tumor patient |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199972A (en) * | 2013-09-22 | 2014-12-10 | 中科嘉速(北京)并行软件有限公司 | Named entity relation extraction and construction method based on deep learning |
| CN104794546A (en) * | 2015-04-29 | 2015-07-22 | 武汉大学 | Wind power climbing forecasting method based on deep confidence network classifying method |
| CN105069825A (en) * | 2015-08-14 | 2015-11-18 | 厦门大学 | Image super resolution reconstruction method based on deep belief network |
| US9846938B2 (en) * | 2015-06-01 | 2017-12-19 | Virtual Radiologic Corporation | Medical evaluation machine learning workflows and processes |
-
2017
- 2017-01-05 CN CN201710007736.5A patent/CN106897545B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199972A (en) * | 2013-09-22 | 2014-12-10 | 中科嘉速(北京)并行软件有限公司 | Named entity relation extraction and construction method based on deep learning |
| CN104794546A (en) * | 2015-04-29 | 2015-07-22 | 武汉大学 | Wind power climbing forecasting method based on deep confidence network classifying method |
| US9846938B2 (en) * | 2015-06-01 | 2017-12-19 | Virtual Radiologic Corporation | Medical evaluation machine learning workflows and processes |
| CN105069825A (en) * | 2015-08-14 | 2015-11-18 | 厦门大学 | Image super resolution reconstruction method based on deep belief network |
Non-Patent Citations (5)
| Title |
|---|
| Computer-aided classification of lung nodules on computed tomography images via deep learning technique;Kai-Lung Hua;《OncoTargets and Therapy》;20150804;正文2015-2022页 |
| Unsupervised Deep Learning Applied to Breast Density Segmentation and Mammographic Risk Scoring;Michiel Kallenberg;《IEEE TRANSACTIONS ON MEDICAL IMAGING》;20160531;第35卷(第5期);正文1322-1331页 |
| 右删失数据风险函数直方图估计;乔舰;《统计与信息论坛》;20150131;第30卷(第1期);正文14-17页 |
| 基于高斯受限玻尔兹曼机的非线性过程故障检测;陈曦;《上海应用技术学院学报》;20150630;第15卷(第2期);正文139-143页 |
| 机器学习在肿瘤早期诊断与预后预测中的应用;施维;《医学信息学杂志》;20161125;第37卷(第11期);正文10-15页 |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111184948A (en) * | 2020-01-09 | 2020-05-22 | 北京理工大学 | Vascular targeted photodynamic therapy-based nevus flammeus treatment method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106897545A (en) | 2017-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106897545B (en) | A kind of tumor prognosis forecasting system based on depth confidence network | |
| CN107622104B (en) | Character image identification and marking method and system | |
| CN104866829B (en) | A kind of across age face verification method based on feature learning | |
| CN110084318B (en) | Image identification method combining convolutional neural network and gradient lifting tree | |
| Mitteroecker et al. | The ontogenetic trajectory of the phenotypic covariance matrix, with examples from craniofacial shape in rats and humans | |
| CN106529503B (en) | A kind of integrated convolutional neural networks face emotion identification method | |
| CN103824054A (en) | Cascaded depth neural network-based face attribute recognition method | |
| CN104462184B (en) | A kind of large-scale data abnormality recognition method based on two-way sampling combination | |
| CN110533024B (en) | Double-quadratic pooling fine-grained image classification method based on multi-scale ROI (region of interest) features | |
| CN109344759A (en) | A kind of relatives' recognition methods based on angle loss neural network | |
| CN109191471A (en) | Based on the pancreatic cell image partition method for improving U-Net network | |
| CN107680600A (en) | Sound-groove model training method, audio recognition method, device, equipment and medium | |
| CN105469034A (en) | Face recognition method based on weighted diagnostic sparseness constraint nonnegative matrix decomposition | |
| CN106970981B (en) | Method for constructing relation extraction model based on transfer matrix | |
| CN103366182B (en) | Based on the full face identification method supervising Non-negative Matrix Factorization | |
| CN108960043A (en) | A kind of personage's family relationship construction method for electron album management | |
| CN111611877A (en) | Age interference resistant face recognition method based on multi-temporal-spatial information fusion | |
| CN108256630A (en) | A kind of over-fitting solution based on low dimensional manifold regularization neural network | |
| CN109857879A (en) | A kind of face retrieval method based on GAN | |
| CN110490028A (en) | Recognition of face network training method, equipment and storage medium based on deep learning | |
| CN106250918B (en) | A kind of mixed Gauss model matching process based on improved soil-shifting distance | |
| Somervuo | Time–frequency warping of spectrograms applied to bird sound analyses | |
| CN106204749A (en) | The threedimensional model of sparse low-rank feature representation is divided into segmentation method | |
| CN108629019B (en) | Question-answer field-oriented question sentence similarity calculation method containing names | |
| CN110503148A (en) | A kind of point cloud object identifying method with scale invariability |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |